economic fundamental 2007

8/11/2019 Economic Fundamental 2007

1/23

ECONOMICFUNDAMENTALSANDEXCHANGERATES 137Journal of Applied Economics. Vol X, No. 1 (May 2007), 137-159

ECONOMIC FUNDAMENTALS AND EXCHANGE RATES

UNDER DIFFERENT EXCHANGE RATE REGIMES:

KOREAN EXPERIENCE

BYUNG-JOO LEE

University of Notre Dame

Submitted February 2005; accepted April 2006

Korea provides a unique opportunity to study the different behaviors or roles, if any, of limited

flexibility and free floating exchange rate regimes. Korea shifted from a limited flexibility to a

free floating exchange rate regime after the 1997 economic crisis. It is well documented that

the exchange rate is very difficult to predict using any theoretical models for exchange rate

determination. Based on a simple monetary model, we find that the impact of economic

fundamentals on the exchange rate is very similar under both exchange rate regimes according

to OLS estimates, but the difference is statistically significant with GARCH(1,1) results. We

also find that the size of the exchange rate shock is much bigger under the free floating regime

than under the limited flexibility regime. VAR results show that the exchange rate shock impact

on inflation is not statistically different under the two regimes. These findings are generally in

line with Baxter and Stockman (1989) for regime neutrality.

JEL classification codes: F31, F43, C22

Key words:Korean exchange rate regimes, economic fundamentals, exchange rate pass-through

I. Introduction

After the recent Asian economic crisis of 1997, many Asian countries, including

Thailand, Malaysia, Indonesia and Korea among others, were forced to devalue

their local currencies and resorted to a free floating exchange rate system. It is

widely believed that fixed or pegged exchange rate regimes are ultimately destined

to collapse, thus resulting in an economic crisis. Obstfeld and Rogoff (1995) and

Larrain and Velasco (2001) argue that the solution to economic crises lies in increased

exchange rate flexibility in the long-run.

Byung-Joo Lee: Department of Economics and Econometrics, University of Notre Dame,

Norte Dame, IN 46556, USA; 574-631-6837; [email protected]. This paper benefited greatly from

the comments by the editor and anonymous referees. All remaining errors are my own.


2/23

JOURNALOFAPPLIEDECONOMICS138

Baxter and Stockman (1989) and Flood and Rose (1995) studied the relationship

between exchange rate regimes and macroeconomic volatilities, and found that

macroeconomic volatilities are not particularly dependent on exchange rate regimes.

This paper follows a similar line of research focusing on a single country, Korea.

This paper investigates the macroeconomic structural differences of the Korean

economy under the free floating exchange rate regime after the economic crisis

compared to the limited flexible regime before the economic crisis. The contribution

of this paper is two fold. First, this paper focuses on Korean exchange rate regimes.The Korean economy has grown so fast in the last 30 years, that Korea is the 12 th

largest economy in the world, joining the OECD in 1996, and becoming one of the

key players in international trade. Even with these achievements, the Korean

economy did not receive due attention in rigorous academic research. Second,

Korean exchange rate regimes provide one of the unique opportunities to study

different behaviors or roles, if any, of a limited flexible regime and a free floating

regime in the same economy. Since the regime change has occurred in a relatively

recent period, it provides a natural experiment to empirically verify the role of

exchange rate regimes on macroeconomic variables. We will investigate the effect

of exchange rate pass-through on domestic variables such as the inflation rateunder the two different regimes. The results of this paper provide useful guidelines

for emerging economies to properly set their exchange rate system for stable

economic growth.

There is an increasing trend for many developing countries to adopt free floating

exchange rate regimes after economic crises. In reality, however, the officially

declared exchange rate regimes are not what they claim to be. This is the fear of

floating by Calvo and Reinhart (2002). The advantages of a fixed regime, especially

for developing countries, are well summarized in Frankel (2003). They are: providing

a nominal anchor to monetary policy, encouraging trade and investment, precluding

competitive depreciation and avoiding speculative bubbles. In short, fixed exchange

regimes provide the stability that developing countries need to maintain their

economic growth. However, as the countries manage to maintain a fixed exchange

rate with occasional interventions, there are usually smaller exchange rate shocks,

but there might be one huge shock that might force the country to abandon the

peg. Under the free floating regime, exchange rates are allowed to move freely to

adjust economics fundamentals, and there may be larger shocks, but there is no


3/23

ECONOMICFUNDAMENTALSANDEXCHANGERATES 139

risk of a massive exchange rate shock. Frankel (2003) also provides four advantages

of a free floating exchange rate regime: independent monetary policy, automatic

adjustment to trade shocks, seigniorage and lender of last resort ability, and ability

to avoid speculative attacks. However, as Frankel points out, it is not completely

clear whether the majority of developing countries can, or are willing to take

advantage of a free floating exchange rate regime.

The next section introduces a simple monetary model of exchange rate

determination based on the purchasing power parity. Section III describes the dataset and presents empirical results. Section IV concludes the paper with some

suggestions on the future direction of the current study.

II. Theoretical framework of exchange rate determination

The theoretical framework of our model is based on the simple monetary model

summarized in Obstfeld and Rogoff (1996). Exchange rate prediction is a notoriously

difficult task, as already demonstrated by Meese and Rogoff (1983). MacDonald

and Taylor (1994), and Mark (1995) use the monetary model to test the predictability

of exchange rates. They claimed the modest success in predicting exchange rates

for a longer horizon. Mark and Sul (2001) use the same model for panel data set of19 industrialized countries, while Wu and Chen (2001) estimated equation (8) below

using a nonlinear Kalman filter allowing for a time-varying nature of the slope

parameter. The monetary model consists of four behavioral equilibrium equations:

domestic and foreign (ROW) money market equilibrium, the purchasing power

parity condition (PPP) and the uncovered interest parity condition (UIP),

,tttt iypm =

,****tttt iypm =

,*ttt pps =

,1*

ttttt ssEii = +

where

( ):*ttmm domestic (foreign) money supply in natural log;):*tt pp domestic (foreign) price level in natural log;

(1)

(2)

(3)

(4)


4/23


):*tt yy domestic (foreign) GDP in natural log;):*tt ii domestic (foreign) interest rate;

:ts nominal exchange rate (local currency price of one foreign currency) in natural

log;

:1+ttsE expectation of 1+ts at time t;

:10


5/23


,101 ttt zs ++= +

where )*ttttt iifsz == is the nominal exchange rate deviations from theeconomic fundamentals. We expect 01


6/23


maintained a managed float throughout the 1980s and 1990s until the 1997 economic

crisis and moved to a free float after the economic crisis and economic restructuring.

However, there are several other classification schemes that do not agree with the

official de jure classification. Reinhart and Rogoff (2004, RR hereafter) using parallel

rates classify the Korean exchange rate regime as a crawling-peg during 80s and

90s and a free float after 1998. Levy-Yeyati and Sturzenegger (2003, 2005, LYS

hereafter), using the volatilities of exchange rates and reserves, classify the Korean

exchange rate regime for the crisis period as an intermediate (crawling peg) regimeand the post-crisis period of 1999 and 2000 as a fixed regime. This classification

does not reflect the true nature of Korean exchange rate regimes in recent years.

Dubas, Lee and Mark (2005, DLM hereafter) use volatilities of effective exchange

rates in addition to the official exchange rates and reserves to classify the Korean

exchange rate regime as a limited flexibility regime (similar to crawling peg) with a

few exceptional years for the 1980s and 1990s until the economic crisis, and

independently floating (free float) after 1998. For the 1980s and 1990s before the

economic crisis, IMF official de jure classification is more flexible than those of RR

and DLM. This paper follows the generally agreed classifications of RR and DLM

for the pre-crisis period as a limited flexibility regime and for the post-crisis periodas a free float.

A. Data description

All our data comes from the IMF International Financial Statistics (IFS) CD-

ROM. Data frequency is monthly except GDP and GDP deflator series, which are

available only on a quarterly basis. We converted the quarterly series into monthly

frequencies by linearly interpolating quarterly observations into monthly

observations.

We used the bilateral nominal exchange rates per U.S. dollars for Australia,

Japan and South Korea for the period of January 1980 to December 2003. These

exchange rates are nominal domestic currency prices per US dollar at the end of

each month. The Japanese yen and Australian dollar are introduced here as

benchmarks for Korean exchange rate regimes. Japan is one of the largest trading

partners of Korea, and Korea has sustained a chronic trade deficit with Japan. In

addition to the close economic relationship between Korean and Japan, the

Japanese yen has been freely floating after the collapse of the Bretton Woods


7/23


Accord. As Calvo and Reinhart (2002) observed, the Japanese yen serves as one

of the reserve currencies of the world. Therefore, its characteristics of free floating

regime may be different from those of small developing economies. In this regard,

Australia is chosen because the Australian dollar is also freely floating, but because

the Australian economy is much smaller than that of Japan, it more closely resembles

typical small developing economies. As in the Calvo and Reinhart (2002) study, we

use the Australian dollar as a benchmark currency for a free floating exchange rate

regime of a small open economy.1

Other economic variables in our analysis are as follows: money supply: M2

measure of nominal money supply; interest rate: short term government bond

rates for Australia and Japan, short term (90 day) deposit rate for Korea, and 3

month U.S. Treasury bill rate; general price level: manufacturing output prices for

Australia, consumer price indices for Japan, Korea and the United States; reserves

are measured as total reserves minus gold in U.S. dollar terms.

We divide our data into three periods according to the regime classifications of

RR and DLM. The first period is from January 1980 to the beginning of the Korean

economic crisis, September 1997 (period 1). During this period, Korean exchange

rates were tightly managed and controlled by the Bank of Korea. Exchange ratechanges are very limited during this period. The second period is the crisis period,

October 1997 to September 1998, when the first round of financial restructuring

was completed following the IMF recommendations to recover from economic

crisis. During the crisis period, nominal exchange rates were unstable and fluctuating

widely. Thus, we exclude this period from our analysis. The last period, starting

October 1998 to the end of sample period, December 2003, is the post crisis free

floating exchange rate regime (period 2). Korean exchange rates were allowed to

move freely during this period with minimal market intervention.2

1Calvo and Reinhart (2002) use the Australian dollar as a benchmark currency for the floating

exchange rate regimes for small developing countries. Australia, with a credible commitment to

floating, shares many characteristics of small developing countries, and the Australian dollar is

not a world reserve currency. Calvo and Reinhart (2002) report that the probability the

Australian dollar fluctuates within the prescribed 2.5% monthly band for a free floating regime

is about 70% during the monthly period of January 1984 to November 1999. For U.S. dollar/

Deutsche mark, this probability is 59% and it is 61% for Japanese yen/U.S. dollar. The average

of those countries declared as floating regimes in Calvo and Reinhart (2002) is over 79%.

2The definition of crisis period could be arbitrary. However, varying the crisis window did not

qualitatively alter our results.


8/23


B. Exchange rates and economic fundamentals

First, we will examine the volatilities of two variables that are closely related to

exchange rate regimes, the nominal exchange rate and foreign reserves. We compare

the monthly rate of return volatilities measured as the standard deviation of the

percentage change of the bilateral nominal exchange rates and foreign reserves

),,( 11 == tttttt rrrsss where st is the natural log of the nominal

exchange rateStand r

tis the natural log of foreign reservesR

t. Table 1 compares the

return volatilities of three exchange rates for two distinct periods, before the Korean

economic crisis for the limited flexibility regime and after the economic crisis for the

free floating regimes. Volatility is measured as the standard deviation of each

variable. This table also provides test statistics to test the equality of the volatilities

between two periods.

Table 1. Volatilities for nominal exchange rates and reserves for each period

Limited flexibility regime, period 1 Free floating regime, period 2

January 1980 September 1997 October 1998 December 2003

ts tr ts trKorea 0.869 7.505 2.629 1.663Australia 2.854 9.477 3.204 7.334

Japan 3.387 3.459 3.481 2.442

F-test statistics for 22,21,0 : iiH =

Korea ts 9.162 (0.000) tr 20.364 (0.000)Australia ts 1.260 (0.308) tr 1.670 (0.011)Japan ts 1.056 (0.830) tr 2.006 (0.001)

Note: Test statistics are for the null hypothesis that volatilities are the same between two periods; p-values

are in parenthesis

Table 1 shows that the Korean won is much less volatile during the limited

flexibility regime, and its volatility is much smaller than that of Australian dollar

and Japanese yen. During the free float regime, the Korean won is still less volatile

than those other exchange rates, but their difference is now statistically


9/23


insignificant.3The Korean wonfluctuates as freely as other floating exchange rate

currencies after adopting the free floating regime in period 2. The Australian dollar

and the Japanese yen show little change in volatility during these two periods

even with the recent Asian economic crisis. Test statistics show little evidence of

volatility changes of the two free floating currencies. The Korean won, on the

other hand, shows strong evidence of volatility change during this period. Korean

foreign reserve holdings are much more volatile under the limited flexibility regime

than under the free floating period. This is expected since under limited flexibilityreserves are often used to maintain stable nominal exchange rates (the interest rate

is another policy tool to manage exchange rates). By comparing the reserve

volatilities of two periods, we can observe that the reserve changes have become

increasingly stabilized under the recent free floating exchange regime, especially

for Korea.

Exchange rate volatilities can be best illustrated using the figures. To avoid

cluttering the figures, Figure 1 plots the nominal exchange rate returns for two

currencies, the Korean won (solid line) and the Japanese yen (broken line) against

the U.S. dollar for the entire sample period. Australian dollar returns could also be

plotted in the same figure, but this is not included in Figure 1 to simplify thepresentation.

The Japanese yen is more volatile during period 1 when the Korean won was

under a limited flexibility regime. During period 2 when the Korean won was freely

floating, the two currency volatilities appear to be quite similar, and they are not

statistically different.

Reserves are often used to control and manage nominal exchange rates under

fixed and limited flexibility exchange rate regimes. Figure 2 plots the volatility of

reserve changes for Korea (solid line) and Japan (broken line). It is clear that

Korean reserves were much more volatile than those of Japan during the limited

flexibility regime. Korean reserves are also more volatile under the limited flexibility

regime than under the free floating regime. This presents the empirical evidence of

exchange rate management schemes. While there are criticisms that Korean

exchange rates are still managed and controlled during the free float period, reserve

3 Levines F-test statistic to test the equality of volatilities of three currencies is 1.410 with p-

value of 0.247. Levines test is based on the analysis of variance (ANOVA) of the absolute

mean difference.


10/23


Figure 1. Monthly percentage changes of the Korean and Japanese nominal exchange rates

per U.S. dollar

1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002-20

-10

0

10

20

30

40KDS

JDS

Figure 2. Monthly percentage changes of the reserves for Korea and Japan

1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002-30

-20

-10

0

10

20

30KDRES

JDRES


11/23


volatility is relatively low. The recent volatility of the Korean nominal exchange

rate exhibits characteristics similar to other free floating exchange rates. In fact,

Korean reserves remain relatively stable and exchange rates are more volatile during

the free floating period. Australia has relatively volatile reserve changes throughout

the period. In fact, even with the free floating exchange rate regime, the probability

reserve changes stay within the 2.5% band is only about 50% according to Calvo

and Reinhart (2002).4 Korean reserve levels were highly volatile during the limited

flexibility regime, but her reserve volatility has decreased significantly under thefree floating regime. Korean reserve volatility is even more stable than Japans

during period 2. From Table 1 statistics, we can see that reserve volatilities have

fallen significantly in period 2 compared to period 1 for all three countries. Since

Korea has changed her exchange regime from period 1 to period 2, the reserve

volatility of Korea has been reduced dramatically.

Once we observed the visual pattern of exchange rates movements, we estimate

equation (8) to examine the relationship between exchange rates and economic

fundamentals. Exchange rates, like many other assets prices, often show the

persistence of volatilities in their evolutions. Table 2 is a summary statistic for the

ARCH residuals for the three currencies.

4 Japan has the highest probability of reserve changes within the 2.5% band, 74%, while the

United States has a probability of 62%.

Table 2. Exchange rate behavior (st): ARCH(1) LM test

Country Korea Australia Japan

Period All Period 1 Period 2 All All

F-statistic 27.415 6.309 0.789 0.368 0.936

(0.000) (0.013) (0.378) (0.544) (0.334)

Asymptotic 2 25.088 6.183 0.806 0.371 0.940(0.000) (0.013) (0.369) (0.543) (0.332)

Note: Null hypothesis is no ARCH residuals. ARCH(1) LM test is performed on the residuals of exchangerate changes fitted on the constant. Different lag lengths of ARCH model produce qualitatively similar results.

p-values are in parenthesis.

The Korean won shows ARCH residuals for period 1 and for the entire period,

while there is no evidence of ARCH residuals during free floating period 2. Even


12/23


though the analysis periods exclude the crisis period, there are several episodes of

ARCH residuals (persistent volatilities) under the limited flexibility regime during

the late 1980s and the middle of 1990s leading to the economic crisis. The Australian

dollar and Japanese yen do not show ARCH residuals either for the entire period

or for two periods separately. It is interesting to observe that ARCH residuals

appear only during the limited flexibility exchange rate regime.5

The following two figures, Figures 3 and 4, show that exchange rates are widely

fluctuating around the deviations from economic fundamentals (ztis standardizedto have zero mean) for Korea and Japan, and it is not an easy task to predict

exchange rates using economic fundamentals. The relationship between exchange

rates and fundamentals for Australia show similar patterns to other countries, but

it is not shown here to conserve space. Meese and Rogoff (1983) have shown that

none of the theoretical exchange rate determination models outperform simple

random walk model in the root mean square criteria. Our objective here is not to

predict the exchange rate using the economic fundamentals, but to investigate the

causal relationship between economic fundamentals and the nominal exchange

rate focusing on the exchange rate regime shift of the Korean won, and to compare

it to other flexible exchange rate regimes.

The basic econometric model to examine the relationship between exchange

rates and economic fundamentals is equation (8) from the monetary model

introduced in section II. Table 3 shows the OLS estimation results for three countries,

using

.101 ttt zs ++= +

In Table 3 the Korean won shows ARCH(1) behavior, so in Table 4 we estimated

equation (8) for Korea with a GARCH(1,1) model using

,211

2110

2 ++= ttt

5 Since the Japanese yen and Australian dollar, two free floating benchmark currencies, do not

show the persistence of volatility during the sample period, the persistence of Korean won

volatility during the limited flexibility regime may be attributed to a sub-optimal (relative to

economic fundamentals) exchange rate policy.

(9)

,1101 ++ ++= ttt

zs (10)

(11)


13/23


Figure 3. Economic fundamentals and the nominal exchange rates for Korea

1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002-20

-10

0

10

20

30

40KZ_ST

KDS

Figure 4. Economic fundamentals and the nominal exchange rates for Japan

1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002-16

-12

-8

-4

0

4

8

12JZ_ST

JDS

where ( )ttt Var = +12 and t is an information set at time t.

OLS results show that the Korean wons fluctuation in response to the

deviations from the economic fundamentals has increase from period 1 to period 2,


14/23


but the difference is not statistically significant. This result is generally in line with

Baxter and Stockman (1989) for the regime neutrality. However, by GARCH(1,1)

estimation, it shows that exchange rate is more sensitive to the economic

fundamentals, and the difference between two periods is statistically significant

(-1.12 vs. -6.23). Since we have observed the ARCH(1) residuals from Table 2,

GARCH estimation is more efficient than OLS results. Japan and Australia also

show the increase of slope estimates between two periods, but their differences

are not statistically significant either. While the difference of individual slopeestimates is statistically insignificant, Chow test for the joint hypothesis of

structural stability between two periods strongly rejected for all three countries.

Table 3. OLS estimation

Korea Australia Japan

Period 1 0

7.637 (1.810)*** -4.825 (4.051) 20.309 (9.746)**

(210) 1

-1.215 (0.296)*** -0.674 (0.589) -1.844 (0.870)**

SSR 146.772 1706.178 2345.673

Period 2 0

37.326 (18.735)** -12.802 (20.301) 55.649 (32.976)*

(55) 1

-6.215 (3.099)** -1.604 (2.526) -5.404 (3.189)*

SSR 358.863 546.258 640.851

Both periods 0

8.882(2.911)*** -2.888 (2.272) 8.392 (5.376)

(265) 1

-1.438(0.477)*** -0.387 (0.318) -0.790 (0.487)

SSR 535.406 2258.068 3042.148

F-statistic 7.684 (0.001) 0.326 (0.722) 2.431 (0.090)

Note: Standard errors in parenthesis. *, **, ***indicate statistical significance at 10%, 5% and 1%, respectively.

F-statistic tests the structural equivalence of the two periods, p-values are in parenthesis.

Table 4. GARCH(1,1) estimation for Korean won

Period 1 Period 2 Both periods

o

7.109 (1.402)*** 37.202(19.352)* 5.279 (1.314)***

o

-1.125 (0.229)*** -6.226 (2.194)* -0.834 (0.215)***

o

0.124 (0.041)*** 0.886 (0.357)** 0.085 (0.026)***

1

0.654 (0.152)*** -0.224 (0.083)*** 0.923 (0.094)***

1

0.341 (0.097)*** 1.082 (0.102)*** 0.400 (0.037)***

Note: Standard errors in parenthesis. *, **,***indicate statistical significance at 10%, 5% and 1%, respectively.


15/23


C. Exchange rates and inflation

We are now ready to investigate the impact of exchange rate pass-through to

domestic economic variables. From the purchasing power parity condition (PPP)

of equation (3), there is a one-to-one relationship between the domestic inflation

rate and the nominal exchange rate assuming constant foreign inflation. Therefore,

we would like to see how the change of the nominal exchange rate affects domestic

inflation rate. Our main focus is the effect of exchange rate regimes on the domesticinflation rate for the Korean economy. An important objective of a fixed exchange

rate regime for a developing economy is to maintain stable price levels to help

increase foreign trade trade. However, aggressive exchange rate defense may

excessively drain foreign reserves, and it may bring further pressure for depreciation

and domestic inflation. The vicious cycle may ultimately result in economic crisis.

We will examine the macroeconomic relationship between inflation and the change

of exchange rates since the 1990s. Figure 5 plots these two variables, inflation

(solid line) and the return of the nominal exchange rate (broken line). Figure 6 is a

scatter gram of these two variables for the two periods. The square symbol

represents the limited flexibility exchange regime (1990:03-1997:09), while the trianglesymbol represents the free floating regime (1998:10-2003:12).

Figure 5. Inflation and the changes of the Korean exchange rate

1990 1992 1994 1996 1998 2000 2002-20

-10

0

10

20

30

40KINFL

KDS


16/23


Figure 6. Scattergram of inflation and exchange rate changes

Inflation rate

0 2 4 6 8 10 12

-6

-4

-2

0

2

4

6

8limited flexibility

free float

Exchange rates have become much more volatile while the inflation rate has

become more stable under the free floating regime than under the limited flexibilityregime. For the statistical regression analysis, I will limit my data to two distinctive

periods of the exchange rate regimes, from March 1990 to September 1997 for the

limited flexibility regime and from October 1998 to December 2003 for the free

floating exchange rate.6 Inflation and exchange rates are analyzed using a bivariate

VAR model focusing on the purchasing power parity of equation (3). The additional

exogenous variables are the percentage change of money supply ( tm ), the realGDP growth rate ( trGDP ), and the Korean import price ( tprimport_ ) as aproxy for the foreign prices.7 Exogenous variables are included as one time-

lagged values. The lag length for endogenous variables is two according to the

Schwarz criteria. Other lag length selections were estimated also, but they did not

change the qualitative relationship between these variables. The inflation rate

6The Korean won was pegged to the U.S. dollar until March 1990.

7 The VAR model is estimated using monthly data. Since GDP data is available only on a

quarterly frequency, real GDP is interpolated using quarterly nominal GDP and GDP deflators.

All other variables, including the left-hand variables of inflation and exchange rate, are at

monthly frequencies.

Exchange

rate

changes

Inflation rate


17/23


appears to show a strong time trend, but Dickey-Fuller tests with a time trend reject

the unit root hypothesis for all periods. A bivariate VAR model may be too simple

to analyze the complete exchange rate determination model introduced in section

II, but the main objective of this analysis is to focus on the pass-through of the

exchange rate to inflation contrasting two different exchange rate regimes.

Therefore, the estimated bivariate VAR model is:

,12 10 ttjjtjjt XYBBY +++= =

where ( ) ,, = ttt InflsY ( )= tttt primportrGDPmX _,, and ( )

= ttt l21 ,are uncorrelated white-noise random shocks. jjBB ,,0 are conformableparameters.

Equation (12) is estimated for the two periods separately, and both periods

combined. The following table presents the estimation results.

Table 5.A shows that the exchange rate is largely unaffected by domestic

inflation, but it is affected by money supply and real GDP growth in period 1. For

period 2, money supply affects inflation significantly, but with the wrong sign.

Table 5.B shows that the one period lagged exchange rate change has a significant

impact on the domestic inflation rate for both periods, but money supply and real

GDP growth do not appear to affect the inflation rate. Import price as a proxy for

the foreign price is statistically significant under the free float regime, while

insignificant under the limited flexibility regime.

The Granger causality test confirms the causal relationship between inflation

and the exchange rate. Table 6 reports Granger causality test statistics with two

lags for each period and the two periods combined. The exchange rate does cause

inflation, but not vice versa. From the VAR estimates, we can infer that one percent

depreciation of the one period lagged Korean won ( 1 ts ) increases the inflation

rate 0.12% for the entire period. This could be a consequence of the small open

economy which is heavily dependent on the intermediate goods imports to promote

exports. Currency depreciation will boost exports, but it also causes worsening

terms of trade, and higher import prices of intermediate goods trigger higher

domestic inflation. It is evident that the export boosting policy during the limited

flexibility exchange rate regime was pursued at the expense of domestic inflation.

However, Table 5.B also shows that the impact of the exchange rate on inflation

(12)


18/23


Table 5. Bivariate VAR model

Period 1 Period 2 Both periods

A. Exchange rate equation

0

0.343 (0.288) 1.596 (1.104) -0.017 (0.367)

Inflt-1

0.038 (0.148) -1.034 (0.756) -0.001 (0.259)

Inflt-2

-0.067 (0.148) 0.643 (0.619) 0.045 (0.254)

st-1 0.144 (0.103) 0.216 (0.144) 0.211 (0.087) **st-2

0.179 (0.105) * -0.104 (0.150) -0.134 (0.088)

mt-1

0.140 (0.056) * * -0.568 (0.298) * -0.045 (0.107)

rGDPt-1

-0.146 (0.070) ** 0.145 (0.179) -0.004 (0.094)

import_prt-1

-0.022 (0.017) 0.032 (0.046) -0.026 (0.019)

Adjusted R2 0.167 0.037 0.029

B. Inflation equation

0

0.364 (0.196)*

0.830 (0.245)***

0.218 (0.115)*

Inflt-1

1.309 (0.101)***

0.894 (0.168)***

1.281 (0.081)***

Inflt-2

- 0.361 (0.101)***

- 0.150 (0.138) - 0.329 (0.080)***

st-1 0.118 (0.070)*

0.106 (0.032)***

0.121 (0.027)***

st-2

- 0.100 (0.071) 0.046 (0.033) 0.014 (0.027)

mt-1

- 0.010 (0.038) - 0.090 (0.066) - 0.009 (0.033)

rGDPt-1

- 0.023 (0.048) - 0.008 (0.040) - 0.021 (0.030)

import_prt-1

- 0.015 (0.011) 0.020 (0.010)**

- 0.002 (0.006)

Adjusted R2 0.940 0.883 0.952

D-F -5.960 (0.000)***

-3.625 (0.036)**

-3.839 (0.017)**

Note: Standard errors in parenthesis. *, **, *** indicate statistical significance at 10%, 5% and 1%,

respectively. D-F is Dickey-Fuller statistics for the inflation rate. p-value is in parenthesis.

does not appear to be statistically different in the two periods. This result is in linewith Baxter and Stockman (1989) that the exchange rate regime is largely neutral to

macroeconomic variables.

Having established the causal relationship between the exchange rate and

inflation, we would like to see the behavior of these variables in response to the

external shocks. Figure 7 and 8 are impulse response functions (IRF) and

accumulated response functions (ARF) for periods 1 and 2 using the Cholesky

decomposition for two years (24 months). Since the Cholesky decomposition is


19/23


8The upper-left corner graphs of IRFs from both figures show the size of the exchange rate

shocks. Observe that two graphs show different scales reflecting the size of shocks for each

period.

Table 6. Granger causality test statistics (F-statistics)

Null hypothesis Period 1 Period 2 Both periods

stdoes not Grange cause Infl

t2.537(0.085) 7.388(0.002) 11.704(0.000)

Inflt does not Grange causes

t0.226(0.799) 0.622(0.541) 0.156(0.856)

Note: Values in parenthesis are p-values for the F-statistic.

sensitive to the order of shocks to the VAR system, we produced two sets response

functions by rotating shock orderings. However, different shock orderings

produced remarkably similar responses to the point of being virtually identical.

We only report the IRF and ARF of shock ordering of ( )., tt InflsFigures 7 and 8 are IRF (upper panel) and ARF (lower panel) for each period.

The lower-left corner of IRF and ARF are responses of inflation to the one standard

deviation shock of the nominal exchange rates. Under the limited flexibility regime

(Figure 7), the impulse responses and accumulated responses are quite mild in

magnitude, and they are statistically insignificant. Under the free-floating regime

(Figure 8), the initial impact on inflation from the exchange rate shock is relativelylarge and statistically significant from two to five months. This shows that under

the managed exchange rate regime, the exchange rate shock does not directly

transmit to other macroeconomic variables, especially the inflation rate. Under the

floating exchange rate regime, the nominal exchange rate depreciation directly

passes through to the domestic price level. However, this difference is largely due

to the different size of the shocks in the two periods.8 Exchange rate shocks are

much bigger under the free float regime than under limited flexibility. Under the

limited flexibility regime, exchange rate policy effectively limited the size of the

exchange rate shocks, and minimized the impact on domestic inflation. The

accumulated impulses remain positive and statistically significant in the long-termunder the free float regime.

The upper-right corner of IRF and ARF are the responses of the exchange rate

to the shocks of inflation. As we demonstrated previously with VAR results and


20/23


Figure 7. IRF and ARF (Period 1) to one standard deviations 2*(s.e.)

-0.2

0. 0

0. 2

0. 4

0. 6

0. 8

1. 0

2 4 6 8 10 12 14 16 18 20 22 24

Response of KDS to KDS

-0.2

0. 0

0. 2

0. 4

0. 6

0. 8

1. 0

2 4 6 8 10 12 14 16 18 20 22 24

Response of KDS to KINFL

-0.4

-0.2

0. 0

0. 2

0. 4

0. 6

0. 8

1. 0

2 4 6 8 10 12 14 16 18 20 22 24

Response of KINFL to KDS

-0.4

-0.2

0. 0

0. 2

0. 4

0. 6

0. 8

1. 0

2 4 6 8 10 12 14 16 18 20 22 24

Response of KINFL to KINFL

-1.5

-1.0

-0.5

0. 0

0. 5

1. 0

1. 5

2. 0

2 4 6 8 10 12 14 16 18 20 22 24

Accumulated Response of KDS to KDS

-1.5

-1.0

-0.5

0. 0

0. 5

1. 0

1. 5

2. 0

2 4 6 8 10 12 14 16 18 20 22 24

Accumulated Response of KDS to KINFL

-4

0

4

8

12

16

2 4 6 8 10 12 14 16 18 20 22 24

Accumulated Response of KINFL to KDS

-4

0

4

8

12

16

2 4 6 8 10 12 14 16 18 20 22 24

Accumulated Response of KINFL to KINFL


21/23


Figure 8. IRF and ARF (Period 2) to one standard deviations 2*(s.e.)

-2

-1

0

1

2

3

4

2 4 6 8 10 12 14 16 18 20 22 24

Response of KDS to KDS

-2

-1

0

1

2

3

4

2 4 6 8 10 12 14 16 18 20 22 24

Response of KDS to KINFL

-.2

.0

.2

.4

.6

.8

2 4 6 8 10 12 14 16 18 20 22 24

Response of KINFL to KDS

-.2

.0

.2

.4

.6

.8

2 4 6 8 10 12 14 16 18 20 22 24

Response of KINFL to KINFL

-3

-2

-1

0

1

2

3

4

5

2 4 6 8 10 12 14 16 18 20 22 24

Accumulated Response of KD S to KDS

-3

-2

-1

0

1

2

3

4

5

2 4 6 8 10 12 14 16 18 20 22 24

Accumulated Response of KDS to KINFL

-0.5

0. 0

0. 5

1. 0

1. 5

2. 0

2. 5

3. 0

2 4 6 8 10 12 14 16 18 20 22 24

Accumulated Response of KINFL to KDS

-0.5

0. 0

0. 5

1. 0

1. 5

2. 0

2. 5

3. 0

2 4 6 8 10 12 14 16 18 20 22 24

Accumulated Response of KINFL to KINFL


22/23


Granger causality tests, the impact of inflation shock on the exchange rates are

statistically insignificant for both periods.

IV. Conclusion

This paper investigated the role of economic fundamentals in the determination

of the exchange rate under different exchange rate regimes. Focusing on the Korean

economy, this paper found that the impact of economic fundamentals on exchangerates has increased under the free float regime. However, the difference between

the two periods is statistically significant only according to the GARCH(1,1)

estimation, not to the OLS estimation. VAR estimation results show that the impact

of exchange rates on domestic inflation is remarkably similar under the two different

regimes. The major difference is that the size of exchange rate shocks has become

much bigger under the free float regime. IRF and ARF show the lasting impact of

the exchange rate shock on a domestic variable, the inflation rate. In short, under

the free floating regime, exchange rates are allowed to move freely, thus reflecting

the underlying economic fundamentals more accurately.

It is true that the exchange rate has become more volatile under the flexibleexchange rate system than under the limited flexibility regime. To reduce short-run

volatility of the flexible exchange rate system, the Korean government needs to

pursue an exchange rate policy to reduce the short-run volatility of the flexible

exchange rate system and promote stable economic growth.

It is still an open question regarding which exchange rate regime is better for

economic growth in the long-run, especially for developing economies. My future

research will expand the current topic to investigate the relationship between

different exchange rate regimes and other macro economic performances, especially

economic growth.

References

Baxter, Marianne and Alan C. Stockman (1989), Business cycles and the exchange-rate

regime: Some international evidence, Journal of Monetary Economics23: 377-400.

Calvo, Guillermo A., and Carmen M. Reinhart (2002), Fear of floating, Quarterly Journal of

Economics , 117: 379-408.

Dubas, Justin, Byung-Joo Lee and Nelson Mark (2005), Effective exchange rate classification

and growth, Working Paper 11272, NBER.


23/23


Flood, Robert P. and Andrew K. Rose (1995), Fixing exchange rates: A virtual quest for

fundamentals,Journal of Monetary Economics, 36: 3-37.

Frankel, Jeffrey A. (2003), Experience of lessons from exchange rate regimes in emerging

economies, Working Paper 10032, NBER.

Larrain, Felipe and Andres Velasco (2001), Exchange rate policy in emerging markets: The

case for floating, Essays in International Economics 224, Princeton, NJ.

Levy-Yeyati, Eduardo and Federico Sturzenegger (2003), To float or to fix: Evidence on the

impact of exchange rate regimes on growth,American Economic Review93: 1173-1193.

Levy-Yeyati, Eduardo and Federico Sturzenegger (2005), Classifying exchange rate regimes:

Deeds vs. words,European Economic Review49: 1603-1635

MacDonald, Ronald, and Mark P. Taylor (1994), The monetary model of exchange rates:

Long-run relationships, short-run dynamics and how to beat a random walk, Journal of

International Money and Finance 13: 276-290.

Mark, Nelson (1995), Exchange rates and fundamentals: Evidence on long-horizon

relationships, American Economic Review85: 201-218.

Mark, Nelson, and Donggyu Sul (2001), Nominal exchange rates and monetary fundamentals:

Evidence from a small post-Bretton Woods panel, Journal of International Economics

53: 29-52.

Meese, Richard A. and Kenneth Rogoff (1983), Empirical exchange rate models of the

seventies: Do they fit out of sample?, Journal of International Economics14: 3-24.

Obstfeld, Maurice, and Kenneth Rogoff (1995), The mirage of fixed exchange rates,Journal

of Economic Perspectives9: 73-96.

Obstfeld, Maurice ,and Kenneth Rogoff (1996), Foundations of International Macroeconomics,

Cambridge, MA, MIT Press.

Reinhart, Carmen M., and Kenneth Rogoff (2004), The modern history of exchange rate

arrangements: A reinterpretation, Quarterly Journal of Economics119: 1-48.

Wu, Jyh-Lin, and Show-Lin Chen (2001), Nominal exchange rate prediction: Evidence from

a nonlinear approach, Journal of International Money and Finance20: 521-532.

economic fundamental 2007

Documents